The goal of this project is to develop a method for rapid, high-resolution diffusion-weighted MRI (DW-MRI) of the brain. Over the last several years DW-MRI has been established as a useful and extremely promising neuroimaging modality. However, because of problems with motion, DW-MRI is typically carried out using single-shot techniques, which are very robust in terms of imaging speed and motion insensitivity, but are fundamentally limited in spatial resolution. As more clinical and scientific applications for DW-MRI arise, and more sophisticated levels of analysis are formulated, the spatial resolutions available from current DW-MRI methodology will become insufficient. The applicants reported having shown that a method combining Radial Acquisition of Data (RAD), with a magnitude filtered back-projection reconstruction, can yield diffusion-weighted images that have high spatial resolution and are extremely insensitive to magnetic susceptibility and patient motion. The major limitation of this method is that it requires long imaging times and can still sensitive tp rotational motion. The applicants propose to develop a rapid version of diffusion-weighted RAD (DIFRAD) where multiple radial lines of Fourier data are acquired each excitation via spin- and/or gradient-echo refocusing. Acquiring multiple radial lines of Fourier data will dramatically increase imaging speed and will also enable more complete correction of motion induced errors. The result of this project will establish a new method for rapid diffusion imaging of the brain with high (sub-millimeter) spatial resolution that is immune to the severe motion artifacts that plague multi-shot DW-MRI. The increased spatial resolution available through DIFRAD methods will significantly enhance the usefulness of DW-MRI as a tool for investigation of normal brain function and anatomy as well as many neuropathologies.